The cross-sectional study was conducted in Harar City, Ethiopia, from 20 June to 10 July 2023. Harar City is situated approximately 526 kilometers from Addis Ababa, with geographical coordinates of 90°20’ N and 42°0′10’ E (20). Harar was the capital city of the Harari region, which comprised six woredas and 19 kebeles, along with nine districts (woredas) encompassing 36 kebeles (21, 22). Since the majority (99.03%) of the population relies on intra-city public vehicles, Harar’s public transportation options include tricycles/Bajajs (n = 9,273), Ladas (n = 189), minibusses (n = 249), and busses (n = 27) (23). Moreover, Harar is a city in Ethiopia’s eastern region, where the country’s trade and business center is located; thus, it is densely populated with people from virtually all ethnicities, professions, and social classes.

The source population for this study included all intra-city public transport vehicles and their drivers operating in Harar City. The study population specifically focused on the intra-city public transport vehicles that provided services in Harar City during the data collection period.

All intra-city public transport vehicles working in Harar City during the study period were included. Vehicles that were undergoing maintenance and not present in the waiting area during this time were excluded.

The sample size for assessing contamination by pathogenic bacteria in intra-town public transport was determined using a single-population proportion formula (n = Z² (1-P) P/d²). With a 95% confidence level and a 5% margin of error, and considering a prevalence rate of pathogenic bacterial contamination on bus surfaces in Mekelle Town (22%) (1), the calculated sample size was 258. The four public transport stations or bus stops were randomly selected from the nine available in Harar City, specifically Aretagna/Iman Masjid, Shell/Ageeb area, Station/Shewa Bar, and Jegol/Feres Magala. The sample size was then distributed proportionally based on the number of seats for each type of public transport in the town, allocating 209 samples for tricycles/Bajaj, 22 for minibusses, 21 for Lada vehicles, and 6 for busses. The swab samples were conveniently collected from various parts of the selected public transport vehicles (Figure 1).

Surface swab specimens were inoculated onto MacConkey agar, blood agar, and 1 in 10 fourfold diluted nutrient agar. The inoculated agar plates were then incubated at 35°C ± 2 for 24 to 48 h. After the incubation period, the growth on the agar plates was examined to isolate and identify the bacteria present (25), and the viable bacterial load in the vehicles was estimated as colony-forming units per square centimeter of surface (CFU/cm²). Bacterial identification is based on colony characteristics, hemolytic appearance, differential properties of the culture media, and Gram reaction. Identification was performed using various standard biochemical tests, including catalase, oxidase, urease, indole, citrate utilization, glucose and lactose fermentation, gas and H2S production, lysine decarboxylation, and motility tests (5, 9, 26).

The antibiotic susceptibility patterns of the identified pathogenic bacterial species were determined using the standard Modified Kirby-Bauer disk agar diffusion method on Mueller Hinton agar, following the guidelines provided by the Clinical and Laboratory Standards Institute (CLSI 2022). A McFarland standard equivalent inoculum of bacteria was swabbed onto the surface of a Mueller Hinton agar plate for antibiotic susceptibility testing. Filter paper disks containing specific antimicrobial agents—ampicillin (10 μg), oxacillin (30 μg), ciprofloxacin (5 μg), gentamicin (10 μg), chloramphenicol (2 μg), ceftriaxone (30 μg), cefuroxime (30 μg), and erythromycin (15 μg)—were placed on the agar surface. Following overnight incubation, the diameter of the zone of inhibition around each disk was measured using the standardized tables provided by CLSI 2022, and a semi-qualitative report of susceptibility or resistance to each antibiotic was determined (27, 28).

The antimicrobial susceptibility pattern reflects how isolated bacteria respond to commonly prescribed drugs, indicating whether they are susceptible or resistant. This pattern is measured by comparing the zone of inhibition against a standard comparison chart, allowing for the interpretation of susceptible, intermediate, and resistant categories (5).

Hygienic practices: these practices refer to the cleaning protocols implemented at each car wash, evaluated with a yes/no response for the following: 1. dry wiping, 2. water only, 3. water and detergents, and 4. water and bleach.

Hospital route: were the vehicles designated as running hospital or non-hospital on the day of sampling (29)?

The sample: It includes the vehicle’s seats, handles (rail), and doors.

The level of pathogenic bacterial isolates: this is determined by counting the colonies and comparing them to WHO standards.

Multiple routes: vehicles were categorized as servicing multiple routes if they served two or more routes on the sampling day, while single-route vehicles served only one route (29).

Multidrug resistance: this is defined as the resistance of an isolate to three or more classes of antibiotics.

A pathogenic bacterial profile: it includes any health-important bacteria. If these bacteria grow on selective media, they are measured as yes/no and confirmed by biochemical tests.

Types of vehicles: there are various types of vehicles, such as: (1) minibusses, (2) Lada/Peugeot, (3) tricycles/Bajaj, and (4) busses; everyone has the right to use them, provided they adhere to the rules governing their use.

Types of surfaces: they include vehicle surfaces such as metals, plastics (such as polyvinyl), and cloth coverings for various vehicle parts.

The questionnaire used in the study underwent a translation process to ensure consistency with the local languages Amharic and Afan Oromo. Initially, professional linguists translated the questionnaire from English to these languages. To validate the accuracy of the translations, the questionnaire was then translated back into English by a different team of linguists. Before the commencement of data collection, the lead investigator trained the data collectors and laboratory personnel involved in the study. This training aimed to familiarize them with the study’s objectives, data collection procedures, and laboratory techniques, ensuring consistency and accuracy in data collection and analysis.

To maintain the sterility of the swabs used for sample collection, rigorous checks were conducted, and swab collection followed aseptic techniques, including the use of sterile tubes for transportation, sterile cotton-tipped sticks for swabbing, and clean, powder-free gloves to minimize contamination. Each collected sample was carefully controlled and recorded. Furthermore, quality control measures were implemented during the Gram staining process. Known control slides were used to validate the quality of reagents and the effectiveness of staining procedures. These controls helped ensure the accuracy and reliability of the Gram staining process. Additionally, the performance of the culture medium and antibiotic disks was evaluated using reference strains of E. coli (ATCC 259), S. aureus (ATCC 25923), and P. aeruginosa (ATCC 27853) (30). The turbidity of the bacterial suspension was adjusted to achieve the 0.5 McFarland level for conducting a drug susceptibility test.

The study received ethical clearance from the Institutional Health Research Ethics Review Committee (IHRERC) at Haramaya University, under reference number IHREC/037/2022. Additionally, the Harari Transport Bureau provided an official letter of support for the research. Prior to sample collection, all participants were thoroughly informed about the study’s objectives and voluntarily provided written consent. Strict measures were implemented to ensure the confidentiality of participant information, and positive results were communicated to the respective vehicle drivers. Comprehensive information about the study, including its objectives, methodology, potential risks, and benefits, was shared with each driver involved. Participants were informed of their right to decline participation or withdraw at any point during the research process.

The total aerobic bacterial load from vehicles (minibusses, Lada taxis, tricycles, and busses) ranged from no growth to 3.05 × 10⁶ CFU/mL, with a mean of 2.994 (SD ± 0.562 × 10⁴ CFU/mL) (Table 2).

The majority of vehicle surface samples were found to be contaminated, with an average bacterial load of 2.994 (SD: 0.562 × 10⁴) CFU/4 cm². Significantly higher mean colony counts were observed on the seat surfaces, at 4.314 × 10⁶ (SD: 0.56 ×10⁶) CFU/4 cm² and door surfaces (p = 0.0035), at 0.415 × 10³ (SD: 0.141 × 10³) CFU/cm² compared to the handles, which had 2.954 × 10⁶ (SD: 0.52 × 10⁶) CFU/cm².

A total of 258 vehicles were sampled, from which pathogenic bacteria were isolated from 192 vehicles, resulting in an isolation rate of 74.4% (95% CI: 69–80%). Additionally, out of the 364 isolates, 297 were identified as pathogenic bacteria, while the remaining isolates were classified as contaminants. The highest isolation rate was observed in Lada vehicles (95.2%), followed by busses (83.3%), minibusses (72.7%), and tricycles (72.2%) (Table 2). Nearly two-thirds, or 240 out of the 364 bacterial isolates, were Gram-positive, while 124 of them were Gram-negative bacteria. Among the Gram-positive isolates, coagulase-negative staphylococci (CoNS) were the most frequently isolated (n = 104), followed by Staphylococcus aureus (n = 71), and 65 were Gram-positive bacilli species. Of the 140 Gram-negative isolates, 30 were Klebsiella species, 28 were E. coli, 23 were Pseudomonas species, 21 were Salmonella species, and 19 were Shigella species. The frequencies of pathogens varied considerably between vehicle types; the bacterial pathogens on bus surfaces included CoNS at 3/6 (50.8%), S. aureus at 3/6 (50.8%), E. coli at 2/6 (33.3%), Klebsiella species at 1/6 (16.7%), Salmonella species at 2/6 (33.3%), Shigella species at 2/6 (33.3%), and Pseudomonas aeruginosa at 1/6 (16.7%) (Figure 2).

The frequencies of bacterial pathogens can also vary considerably between different types of vehicle surfaces. Among the 297 pathogenic bacteria isolated, 151 (50.84%) were found on seats, 110 (37%) on handles, and 36 (12.12%) on doors. Regarding the seats, the breakdown is as follows: CoNS 49 (55.7%), S. aureus 19 (27.8%), E. coli 18 (18.6%), Klebsiella spp. 18 (18.6%), Salmonella spp. 11 (11.3%), Shigella spp. 10 (10.3%), and Pseudomonas aeruginosa 13 (13.4%).

A total of 297 pathogenic bacterial isolates were tested against eight different antibiotic disks, revealing higher proportions of resistance rates among Gram-positive bacteria to ampicillin (97.6%) and oxacillin (56.5%). The resistance rate of S. aureus to ampicillin (97.6%) and oxacillin (64.1%) is greater than that of CoNS, which shows resistance rates of 95.3% to ampicillin and 40.0% to oxacillin. In contrast, CoNS exhibits a higher resistance rate to cefotaxime (44.3%) compared to S. aureus (28.2%).

Escherichia coli was found to be the most resistant Gram-negative bacterium among the investigated antibiotics: ceftriaxone 64.3% (18/28), ampicillin 94.9% (26/28), chloramphenicol 14.3% (4/28), and cefotaxime 35.7% (10/28). Klebsiella species exhibited the highest resistance rates, including ampicillin (100%), ceftriaxone (73.3%), ciprofloxacin (6.6%), cefotaxime (53.3%), and chloramphenicol (26.7%). Salmonella species showed resistance to chloramphenicol at 28.7% (6), 61.9% to ceftriaxone, and 38.1% to cefotaxime. Shigella species exhibited resistance to ceftriaxone (61.9%), cefotaxime (38.1%), and chloramphenicol (28.7%). The resistance rate in Pseudomonas spp. was 100% to ampicillin, 54.5% to ceftriaxone, 41% to cefotaxime, and 38.1% to chloramphenicol (Table 3).

The antibiogram of Gram-positive bacterial isolates (15.6%) did not show resistance to any of the seven antibiotic classes tested (ampicillin, oxacillin, erythromycin, chloramphenicol, gentamicin, ceftriaxone, ciprofloxacin, and cefotaxime); however, none of the isolates exhibited resistance to all the antibiotics tested. MDR was observed in 67 out of 297 (22.5%) isolates. Notably, a higher rate of MDR was detected in 34 out of 121 (28.1%) Gram-negative isolates. More frequently, 53.6% of E. coli were multidrug-resistant, followed by Klebsiella species at 43.3%. Overall, among the total isolates (n = 67/297), multidrug resistance, defined as resistance to ≥3 antibiotic classes, is shown in Figure 3.

A multiple linear regression analysis was conducted to determine whether the independent variables are related to bacterial load. Handles: (β = 1.18, t = 6.602, p < 0.001), seats: (β = 1.65, t = 9.56, p < 0.001), cloth surfaces: (β = 0.433, t = 2.451, p < 0.015), multi-route services: (β = 0.496, t = 3.36, p < 0.001), more than 300 passengers per day: (β = 0.57, t = 4.036, p < 0.001), and regularly not cleaned: (β = 0.497, t = 3.521, p < 0.001) are positively associated with bacterial load. This suggests that tricycles, handles, the number of passengers per day ≥300, regularly not cleaned, multi-route services, cloth surfaces, and seats play a pivotal role in predicting CFU. Meanwhile, the results also indicate that busses: (β = 0.876, t = 1.83, p < 0.069), Lada (β = 0.367, t = 1.13, p < 0.259), plastic surfaces (β = 0.083, t = 0.51, p < 0.610), and hospital routes (β = 0.055, t = 0.400, p < 0.690) do not impact bacterial loads (Table 4).

There was a statistically significant difference in the multiple routes (p < 0.001), the number of passengers greater than 300 per day (p < 0.001), and those that are regularly uncleaned (p < 0.001). The tolerance (0.911) and variance inflation factor (2.057) values did not indicate a violation of this assumption. The Durbin–Watson statistic was calculated to assess the assumption that the value of the residuals is independent, suggesting that this assumption was not violated (1.17).

In the bivariate analysis, factors such as marital status, type of taxi (vehicle), sample location, presence of multiple routes, hospital routes, daily passenger count, regular cleaning practices, and sample collection time were considered for multivariable logistic regression with a p-value of <0.25. The subsequent multivariable logistic regression analysis identified several significant variables associated with pathogenic bacteria contamination of vehicle surfaces, including sample locations on handles and seats, the presence of multiple routes, hospital routes, a daily passenger count exceeding 300, regular uncleanliness, and sample collection time in the afternoon, all with p-values <0.05.

The multivariable analysis revealed that sample locations from vehicle handles were 2.5 times more likely to be contaminated with pathogenic bacteria compared to door surfaces (AOR: 2.5; 95% CI: 1.1, 6.2). Seats exhibited a five-fold increase in contamination by pathogenic bacteria compared to door surfaces (AOR: 5; 95% CI: 2.1, 12). Vehicles operating on multiple routes had contamination levels from pathogenic bacteria that were 3.7 times higher than those that did not (AOR: 3.7; 95% CI: 1.4, 9.7), while vehicles on hospital routes were 2.36 times more likely to be contaminated by pathogenic bacteria compared to those that were not (AOR: 2.36; 95% CI: 1.1, 5.1). Furthermore, vehicles accommodating more than 300 passengers per day exhibited 2.2 times greater contamination than those with fewer passengers (AOR: 2.2; 95% CI: 1.2, 4.5). Regularly uncleaned vehicle surfaces were 3.6 times more contaminated than clean surfaces (AOR: 3.6; 95% CI: 1.7, 7.3). In addition, vehicles sampled in the afternoon were 3.5 times more likely to be contaminated than those sampled in the morning (AOR: 3.5; 95% CI: 1.7, 7). These findings highlight the significant associations between vehicle characteristics, usage patterns, and the risk of pathogenic bacterial contamination (Table 5).